Meta-llama
Meta-llama
vs
Meta-llama
Meta-llama

Llama 3.1 70B Instruct vs Llama 4 Maverick

A detailed comparison of pricing, benchmarks, and capabilities

Get our weekly newsletter on pricing changes, new releases, and tools.

Key Takeaways

Llama 3.1 70B Instruct wins:

  • Cheaper output tokens
  • Higher intelligence benchmark

Llama 4 Maverick wins:

  • Cheaper input tokens
  • Larger context window
  • Supports vision
Price Advantage
Llama 3.1 70B Instruct
Benchmark Advantage
Llama 3.1 70B Instruct
Context Window
Llama 4 Maverick
Speed
N/A

Pricing Comparison

Benchmark Comparison

Context & Performance

Capabilities

Feature Comparison

FeatureLlama 3.1 70B InstructLlama 4 Maverick
Vision (Image Input)
Tool/Function Calls
Reasoning Mode
Audio Input
Audio Output
PDF Input
Prompt Caching
Web Search

License & Release

PropertyLlama 3.1 70B InstructLlama 4 Maverick
LicenseOpen SourceOpen Source
AuthorMeta-llamaMeta-llama
ReleasedJul 2024Apr 2025

Llama 3.1 70B Instruct Modalities

Input
text
Output
text

Llama 4 Maverick Modalities

Input
textimage
Output
text

Frequently Asked Questions

Llama 4 Maverick has cheaper input pricing at $0.20/M tokens. Llama 3.1 70B Instruct has cheaper output pricing at $0.39/M tokens.
Llama 3.1 70B Instruct has a 131,072 token context window, while Llama 4 Maverick has a 1,048,576 token context window.
Llama 3.1 70B Instruct does not support vision. Llama 4 Maverick supports vision.